Is there anything I can do with lazy GPUs, I have 1 that’s always 1-2 hash slower then the others. Every time I try tweeking, the rates always go down
I did the same as above, flashing my Reference RX 6800 to the XT Version and then went back to the original bios, i don’t know why it works but it does, now the SoC Clock is locked at 960 which is what we wanted so now we can run at full hashrate
This is what I did originally, but deleted my post. I decided to just stick with the 6800 xt bios. With the orginal bios flashed back onto the cards your power usage shows 100-110 but the power at the wall is still 130-140. On top of no power gains, one of the time’s i restarted my rig the soc limit reverted back and my rig became unstable.
True. same issue. Do not stay after reboot. Need to look what we can do with the XT bios
in my case the soc doesn’t reset, even after several reboots i got soc at 960 Mhz, for now this is the most stable i got with accurate power consumption (checking with powermeter)
increasing memory clock causes instability
My soc is at 960 too . But it was at 1200 before reboot.
=== GPU 0, 0d:00.0 Radeon RX 6800 16368 MB ===
Bios: 113-D4120900-101
Core: 1350 MHz 637mV, Mem: 1064 MHz
PerfCtrl: manual, Load: 99%, MemLoad: 100%, Power: 99.0 W, Cap: 203 W
Core: 51°C, HotSpot: 56°C, Mem: 66°C, Fan: 100%, RPM: 3300
Core state: 1, clocks: 500 1350*
Mem state: 3, clocks: 96 456 673 1064*
SOC state: 1, clocks: 800 960*
DCEF state: 1, clocks: 417 738* 1200
F state: 1, clocks: 500 955* 1551
PCIE Link speed: n/a, PCIE Link width: n/a
Memory total: 16368.00 MB, used: 4628.16 MB, free: 11739.84 MB, type: Samsung GDDR6
VDDCR_SOC: 1025mV, VDDCR_CPU: 1056mV
[2021-08-16 01:36:27] ------------------------------------------------------------------------------------------------
[2021-08-16 01:36:27] Mining ethash with 1 GPU workers
[2021-08-16 01:36:27] GPU PCIe CUs CoreMHz SocMHz MemMHz TEdge TJct TMem FanPct FanRpm VDDC Power ETH Cfg
[2021-08-16 01:36:27] 0 0d:00.0 60 1350 960 1064 51C 56C 64C 100.00% 3300 643 mV 98 W A432
[2021-08-16 01:36:27]
[2021-08-16 01:36:27] Stats Uptime: 0 days, 05:22:01
[2021-08-16 01:36:27] ----------------------------------------- GPU Status -------------------------------------------
[2021-08-16 01:36:27] GPU 0 [51C, fan 99%] ethash: 58.20Mh/s, avg 57.96Mh/s, pool 56.16Mh/s a:868 r:0 hw:0
[2021-08-16 01:36:27] Total ethash: 58.20Mh/s, avg 57.96Mh/s, pool 56.16Mh/s a:868 r:0 hw:0
[2021-08-16 01:36:27] ----------------------------------------- Pool Status ------------------------------------------
[2021-08-16 01:36:27] eu-eth.hiveon.net ethash: 58.20Mh/s, avg 57.38Mh/s, pool 56.16Mh/s a:868 r:0
[2021-08-16 01:36:27] eu-eth.hiveon.net ethash: 0.000 h/s, avg 0.000 h/s, pool 0.000 h/s a:0 r:0
[2021-08-16 01:36:27] use-eth.hiveon.net ethash: 0.000 h/s, avg 0.000 h/s, pool 0.000 h/s a:0 r:0
[2021-08-16 01:36:27] use-eth.hiveon.net ethash: 0.000 h/s, avg 0.000 h/s, pool 0.000 h/s a:0 r:0
[2021-08-16 01:36:27] ------------------------------------------------------------------------------------------------
After several tries I managed to find the best performance like this.
I changed the miner too.
This helped me a lot, it was difficult to find a configuration for PowerColor boards
I was having a lot of issues with my reference AMD 6800 that worked great in Windows (between 63 and 64) but then dropped to 58 under HiveOS. After reading this thread, what fixed it for me was applying the 6800 XT BIOS but then my power was way too high. I think applied back the stock 6800 bios (saved my own before modding) and power came down and hashrate stayed.
I’m now at a consistent 63 MH/s with about 114 watt power draw. I haven’t tweaked much since this but right now i feel its pretty stable.
Until you restart or ur rig crash 
Sometime your soc will be 800 / 960 / 1200 completely random and your hash rate will vary
I think we missing something 
So flash to XT is a no go?
Sounds good. Think I will give it a try. I am only getting 47 ATM.
flashed, power usage shot up to 200 watts, so I am flashing back in the original.
So similar behavior after flashing back to stock. SOC stayed at 960. But the core MHz was not following what was input from the Hive interface. So rebooted again. The card is now running at 62MHs, and about 142 w. I will let it run and watch it. hope it stays. wish I could get the core voltage down. There has got to be a better way to modify these BIOs.
I don’t know why I get soc at 1200. it’s matbe why I’m getting arround 100/110 power ?
maybe try to lower your vcore to 650 like me ?
Thought I would share my info. PowerColor Red Dragon 6800s stock!
Unfortunately I can’t run PhoenixMiner and had to settle for TRM
1 Like
This is pretty good for linux.
You’re currently at 0.549 efficiency. Anything above 0.500 efficiency is pretty good for eth mining.
Windows can do 0.560 efficiency but will never be as stable as HiveOS. I have a feeling that you might get a higher hashrate with PhoenixMiner but that might mean a long night of trying to get the OC settings work.
Thanks for the response. I agree with the windows being a pain. With windows my computer would never see my 8th GPU and was really really slow. Thats why I whiched over to linux. It’s been so steady it crazy lol.
After a lot of bios flash…
1.) Flashing XT bios now give me 25mhs…
2.) Revert to original bios give me no control on the card I mean I can’t control the OC througt hive OS
When I look AMD-INFO. core is a 2300… memory 1000. at the end my card is hashing at 58mhz
for 140W ~180W …
Anyone can help me ?
I can only setup the fan
ERROR log after applying settings
Sun Aug 22 14:31:11 EEST 2021
Detected 1 AMD cards
GPU BUS ID : 09
CORE_CLOCK :
CORE_VDDC :
CORE_STATE :
MEM_CLOCK :
MEM_STATE :
MVDD :
VDDCI :
SOCCLK :
SOCVDDMAX :
REF :
FAN :
PL :
AGGRESSIVE ==== GPU 0, 09:00.0 Radeon RX 6800 16368 MB #3 === 14:31:11
ERROR: Invalid parameter “sHeader”, available ones are: header, table_revision, table_size, golden_pp_id, golden_revision, format_id, platform_caps, thermal_controller_type, small_power_limit1, small_power_limit2, boost_power_limit, software_shutdown_temp, reserve, power_saving_clock, overdrive_table, smc_pptable
ERROR: Incorrect variable path: sHeader/TableFormatRevision
ERROR: Invalid parameter “MclkDependencyTable”, available ones are: header, table_revision, table_size, golden_pp_id, golden_revision, format_id, platform_caps, thermal_controller_type, small_power_limit1, small_power_limit2, boost_power_limit, software_shutdown_temp, reserve, power_saving_clock, overdrive_table, smc_pptable
ERROR: Incorrect variable path: MclkDependencyTable/NumEntries
ERROR: Invalid parameter “SclkDependencyTable”, available ones are: header, table_revision, table_size, golden_pp_id, golden_revision, format_id, platform_caps, thermal_controller_type, small_power_limit1, small_power_limit2, boost_power_limit, software_shutdown_temp, reserve, power_saving_clock, overdrive_table, smc_pptable
ERROR: Incorrect variable path: SclkDependencyTable/NumEntries
ERROR: Invalid parameter “MaxODEngineClock”, available ones are: header, table_revision, table_size, golden_pp_id, golden_revision, format_id, platform_caps, thermal_controller_type, small_power_limit1, small_power_limit2, boost_power_limit, software_shutdown_temp, reserve, power_saving_clock, overdrive_table, smc_pptable
ERROR: Incorrect variable path: MaxODEngineClock
ERROR: Invalid parameter “MaxODMemoryClock”, available ones are: header, table_revision, table_size, golden_pp_id, golden_revision, format_id, platform_caps, thermal_controller_type, small_power_limit1, small_power_limit2, boost_power_limit, software_shutdown_temp, reserve, power_saving_clock, overdrive_table, smc_pptable
ERROR: Incorrect variable path: MaxODMemoryClock
ERROR: Invalid parameter “PowerTuneTable”, available ones are: header, table_revision, table_size, golden_pp_id, golden_revision, format_id, platform_caps, thermal_controller_type, small_power_limit1, small_power_limit2, boost_power_limit, software_shutdown_temp, reserve, power_saving_clock, overdrive_table, smc_pptable
ERROR: Incorrect variable path: PowerTuneTable/TDP
ERROR: Invalid parameter “PowerTuneTable”, available ones are: header, table_revision, table_size, golden_pp_id, golden_revision, format_id, platform_caps, thermal_controller_type, small_power_limit1, small_power_limit2, boost_power_limit, software_shutdown_temp, reserve, power_saving_clock, overdrive_table, smc_pptable
ERROR: Incorrect variable path: PowerTuneTable/TDC
ERROR: Invalid parameter “PowerTuneTable”, available ones are: header, table_revision, table_size, golden_pp_id, golden_revision, format_id, platform_caps, thermal_controller_type, small_power_limit1, small_power_limit2, boost_power_limit, software_shutdown_temp, reserve, power_saving_clock, overdrive_table, smc_pptable
ERROR: Incorrect variable path: PowerTuneTable/MaximumPowerDeliveryLimit
ERROR: Invalid parameter “VddcLookupTable”, available ones are: header, table_revision, table_size, golden_pp_id, golden_revision, format_id, platform_caps, thermal_controller_type, small_power_limit1, small_power_limit2, boost_power_limit, software_shutdown_temp, reserve, power_saving_clock, overdrive_table, smc_pptable
ERROR: Incorrect variable path: VddcLookupTable/NumEntries
Legacy mode: voltage and clocks can be changed only via bios
Max memory state: 3, Max core state: 2
Setting DPM to auto mode
Maybe you must reinstall HiveOS. I think there is something wrong with power tables. If you have windows you can test where is the problem GPU Bios or HiveOS? I also revert back my GPU’S to original bios and now they still work at SOC 1200Mhz after rebooting and updates.